Cross-sectional moment of inertia (CSMI) and volumetric cortical bone mineral density (vCtBMD) were assessed by peripheral quantitative computed tomography (pQCT) at femur midshafts from 103 Wistar female rats receiving 0 (n = 12) or 15-1000 mu g/kg/day sc of dexamethasone (n = 46) from 5 to 9 weeks of age, or 0 or 80 mg/kg 3/wk of AI(OH)(3) IP (n = 23,22) from 4 to 10 months of age. A bone strength index (BSI), calculated as the product CSMI x vCtBMD, was found to closely correlate (r = 0.94, R(2) = 0.89, p < 0.001) with the actual, mechanically tested bending breaking force of all bones. Correlation and determination coefficients obtained were higher than those usually reported employing different long-bone strength predictive formulae. The curve approached the origin and was linear throughout the wide range of CSMI, vCtBMD and BSI achieved because of age- and treatment-induced differences, showing a very low standard error of the estimate. Instead, different curve slopes and/or intercepts were found in separate analysis between data from each of the experiments when breaking force was correlated with CSMI or vCtBMD alone, or with the DEXA-assessed BMD of the mechanically assayed bone portion. Results suggest that noninvasive assessment of the BSI by means of pQCT technology provides an original tool for a precise and accurate estimation of long-bone bending strength that can be advantageously applied in crosssectional as well as longitudinal, in vivo studies employing animal models.